Integrating with a targets pipeline

For projects using the targets package, rendering the Quarto document should be the final step in the pipeline. We recommend using tarchetypes::tar_quarto() for this rather than wrapping a render call in a plain targets::tar_target().

Why tarchetypes::tar_quarto() instead of targets::tar_target()?

You could imagine that one approach to render a quarto document in a targets pipeline would be to define a target like this:

```{r}
# Not recommended
targets::tar_target(report, quarto::quarto_render("analysis/final_report.qmd"))
```

However, the problem is that targets has no visibility into what the .qmd file actually depends on. It treats the render as a black box, so it cannot detect when the document’s upstream targets have changed and the report needs to be re-rendered. You end up either re-rendering on every targets::tar_make() call or, worse, serving a stale report without realizing it.

tarchetypes::tar_quarto() solves this by analyzing your .qmd file for targets::tar_read() and targets::tar_load() calls and automatically registering those targets as dependencies. This means targets actually re-render the report when something the quarto document actually reads from the pipeline has changed — no manual dependency tracking required.

```{r}
# Recommended
tarchetypes::tar_quarto(
  name = quarto_notebook,
  path = "qmd/quarto_notebook.qmd",
  quiet = FALSE
)
```

Inside your .qmd file, use targets::tar_read(target_name) or targets::tar_load(target_name)to pull results from the pipeline. These calls are what tar_quarto() scans to build the dependency graph.

If you have a strong reason to use targets::tar_target() directly (e.g., a non-standard render workflow), it is still possible, but you will need to manually declare dependencies using inside the target expression to ensure correct invalidation behavior. We strongly recommend simply using tarchetypes::tar_quarto instead.